The low-density lipoprotein receptor domain class A, or complement-type repeat (CR), constitutes a large family of conserved protein sequences. Structural data on members of this family suggests that CR sequences adopt a characteristic fold, the LDL receptor-like module (Structural Classification of Proteins, SCOP, terminology). CR sequences are found in a variety of different types of proteins including the low density lipoprotein receptor family and the type II transmembrane serine protease (matriptase) family. The LDLR are a family of cell-surface, transmembrane proteins that mediate a wide variety of physiological phenomena. Mechanisms of action include both trafficking of bound ligands and signal transduction from the extracellular space (1). The LDLR participate in various cellular functions, including but not limited to the metabolism of lipoproteins (2, 3), control of matrix metalloproteases and coagulation factors (4-6), specification of cell fate (3, 7), guidance of neural cell migration (7, 8), induction of proliferation in tumor cells (9, 10), binding of rhinovirus (11, 12), signalling by neurotransmitters (13, 14), acquisition of antigens by antigen presenting cells (15), trancytosis of ligands across the blood-brain barrier (16-19), recovery of proteins from glomerular filtrate (20), transport of endocrine hormones (21), efflux of amyloid β peptide from the brain (22), activation of bone deposition (23) and regulation of endothelial cell proliferation (24). The capacity of the LDLR to serve in so many roles derives in part from the diverse set of ligands to which these receptors are able to bind. Another feature of this receptor family is the diverse, and often unique, tissue distribution patterns of each LDLR. The type II transmembrane serine protease family includes corin and the matriptases ST14, matriptase-2 and matriptase-3. Matriptase (MT-SP1, ST14, TADG-15) is overexpressed in a variety of epithelial tumors (carcinomas) (25-33). Following transactivation induced by hepatocyte activator inhibitor-1 (HAI-1), matriptase promotes tumor growth and metastasis by degrading extracellular matrix components directly or by activating other proteases, such as urokinase plasminogen activator (uPA), resulting in matrix-degradative events (26, 34, 35). In addition to the LDLR and matriptase families, a variety of other proteins have CR domains. One such protein, the FDC-8D6 antigen (CD320) has a pair of such domains and plays an important role in B-cell differentiation in lymphatic follicles (36, 37).
The important roles that CR-containing proteins play in pathophysiological processes, along with the unique tissue-distribution profiles of some members of these families, make these proteins useful drug targets. Protein-selective drugs could directly impact the function of a targeted protein, diminishing the supporting effects that the protein has on a particular disease state. Alternatively, the drug could take advantage of the tissue distribution of the targeted protein to efficiently deliver other therapeutic molecules to a particular tissue affected by a disease. Despite considerable evidence of the importance of CR-containing proteins in mammalian physiology and pathophysiology, there are few examples of drugs that act selectively on particular members of the LDLR or CR-containing protein families. The ability to create molecules that bind specific members of these families would provide a means of developing such drugs.
Thus, there exists a need for agents with improved ability to bind selectively to specific receptors including but not limited to members of the LDLR and type II transmembrane serine protease families, either to directly affect the behavior of these CR-containing proteins through the selective binding event or to facilitate delivery of a therapeutic agent to its site of action through the selective binding event.
One example where such receptor-selective molecules might be valuable is in the delivery of therapeutic molecules to the brain. There are approximately 4.5 million people in the US suffering from Alzheimer's disease, and another million with Parkinson's disease. Protein therapeutics for both disorders have shown promise in pre-clinical trials, but obstacles related to drug delivery have slowed development of these drugs (38, 39). The blood-brain barrier (BBB) is a physical and metabolic barrier that separates the peripheral circulation from the central nervous system (CNS). While the BBB serves to protect the microenvironment of the brain, it also presents a challenge to the delivery of therapeutic drugs to the CNS (40, 41). Vehicle-mediated delivery has been widely explored as a means of protein-based drug delivery, but progress to date has been limited. Megalin is expressed on the BBB and there is evidence suggesting that this receptor can mediate transport of ligands into the brain. The best characterized ligands for megalin, including RAP and a variety of lipoproteins, may also exhibit nonspecific binding to other LDLR or are present at saturating levels in blood, competing with the binding of therapeutics. More selective ligands for megalin, molecules with no significant competition from endogenous ligands, would have improved trans-barrier transport properties.
Similarly, the VLDLR has been shown to be expressed on brain capillary endothelium and to mediate transport of lipoprotein lipase across the endothelium of the aorta (Wyne, et al. (1996) Arterioscler Thromb Vasc Biol 16, 407-415; Obunike, et al. (2001) J Biol Chem 276, 8934-8941). Molecules that selectively bind to VLDLR or fusions containing drugs conjugated to such molecules might be expected to have enhanced distribution to the brain. There are potential applications for VLDLR-selective agents outside of brain delivery. The VLDLR has also been implicated in foam cell formation by mediating uptake of excess free fatty acids (FFA) into vascular macrophages (Hiltunen, et al., (1998) Circulation 97, 1079-1086; Qu, et al., (2004) J Huazhong Univ Sci Technolog Med Sci 24, 1-4, 8). Molecules that selectively bind to VLDLR might be developed to block association of lipoprotein particles with macrophages and inhibit foam cell formation. Such molecules would also be expected to limit transfer of FFA from circulating lipoprotein into adipocytes, slowing the progression toward obesity in susceptible subjects (Goudriaan, et al. (2001) Arterioscler Thromb Vasc Biol 21, 1488-1493; Goudriaan, et al. (2004) J Lipid Res 45, 1475-1481; Tacken, et al. (2001) Curr Opin Lipidol 12, 275-279; Yagyu, et al., (2002) J Biol Chem 277, 10037-10043). The high level of expression of VLDLR on the luminal surface of muscle endothelium, along with the low level of expression of VLDLR in liver, would be expected to drive distribution of VLDLR-selective RAP variants to muscle tissue after intravenous administration. Molecules with therapeutic effects on muscle tissue could be attached to VLDLR-selective agents to improve distribution of such molecules to muscle.
Treatments for other diseases may also be developed using ligands with enhanced selectivity for particular CR domains or combinations of CR domains. Overexpression of at least two LDLR, LRP5, LRP6, as well as matriptase (MT-ST1, ST14, TADG-15), has been associated with increased tumorigenicity of the affected tissue (Li, et al., (2004) Oncogene 23, 9129-9135; Hoang, et al., (2004) Int J Cancer 109, 106-111; Tanimoto, et al., (2005) Br J Cancer 92, 278-283; Santin, et al., (2004) Int J Cancer 112, 14-25; Santin, et al., (2003) Cancer 98, 1898-1904; Tanimoto, et al., (2001) Tumour Biol 22, 104-114). Molecules that bind to these proteins may provide a means of diminishing their tumorigenic effects, especially by interfering with their functions directly or but also potentially by targeting tissues that overexpress these proteins with anti-tumor drugs attached to the selective molecule. Matriptase is anchored in the lateral or basolateral membranes of epithelial cells through an N-terminal type II transmembrane domain (42). The membrane-embedded sequence is followed by an extracellular SEA domain, two CUB domains, four CR domains and a trypsin domain at the C-terminus of the protein. Mutagenesis of the CR sequences within matriptase results in a failure of the resulting protease mutant to become activated (43). Similarly, an antibody that binds to the third CR domain of matriptase blocks activation of the enzyme (44). A RAP variant with affinity for one of the two CR pairs within matriptase that include the third CR domain would be expected to interfere with proteolytic activation, in a manner similar to the observed inhibition by the antibody to this region. Such a variant would be expected to diminish the metastatic and tumorigenic effects of matriptase overexpression in affected tissues.
There are approximately eight million women in the US with osteoporosis. Enhanced Wnt signaling through the LDLR LRP5 has been demonstrated to increase osteoblast differentiation, inhibit osteoclast activity and enhance bone deposition (Westendorf, et al. (2004) Gene 341, 19-39; Zhang, et al., (2004) Mol Cell Biol 24, 4677-4684; Mizuguchi, et al. (2004) J Hum Genet. 49, 80-86). This mechanism has been validated with osteoblast-specific APC (adenomatous polyposis coli) knockout mice and with LRP5 mutants that are insensitive to DKK (Dickkopf)-1 and sclerostin-mediated inhibition (Zhang, et al., (2004) Mol Cell Biol 24, 4677-4684; Holmen, et al. (2005) J Biol Chem). Intravenously-administered molecules that selectively bind LRP5 and interfere with inhibitor binding to LRP5 or otherwise enhance Wnt signaling through LRP5, might counter the effects of osteoporosis. Such molecules are not currently available.
Non-Hodgkin's lymphoma (NHL) involves the proliferation and extranodal migration of a class of immune cells called B-cells. NHL is the leading cause of death from cancer in males between the ages of 20 and 39. Studies have shown that the FDC-8D6 antigen protein (CD320) facilitates neoplastic B-cell growth (36, 37). 8D6 antigen contains a single pair of CR domains. Agents, such as RAP variants, that bind to and block the function of 8D6 antigen might then be expected to slow the progression of non-Hodgkin's lymphoma in humans.
Given the widespread participation of CR-containing proteins throughout mammalian physiology, polypeptides with binding-selectivity for particular CR domains have many possible pharmaceutical applications.